Kuntpong Woraratpanya

Improved Triangle Box-Counting Method for Fractal Dimension Estimation

A fractal dimension (FD) is an effective feature, which characterizes roughness and self-similarity of complex objects. However, the FD in nature scene requires the effective method for estimation. The existing methods focus on the improvement of selecting the suitable height of box-counts. This cannot overcome the overcounting problem, which is a key factor to have an impact on the accuracy of the FD estimation. This paper proposes a more accurate FD estimation, an improved triangle box-counting method, to increase the precision of box-counts associated with box sizes. The triangle-box-partition technique provides the double precision for box-counts, thus it can solve the overcounting issue and enhance the accuracy of the FD estimation. The proposed method is evaluated its performance in terms of fitting error. The experimental results show that the proposed method outperforms the existing methods, including differential box-counting (DBC), improved DBC (IDBC), and box-counting with adaptable box height (ADBC) methods.

Improved Thai text detection from natural scenes

Thai text detection from natural scenes is still a challenging task for language translation applications, since there are many unsolved issues. Furthermore, the existing related works cannot completely detect Thai text. The main reason is that Thai text layout has vowels and tonal marks that differ from other languages. This paper proposes an approach to detect Thai text from natural scenes. The approach consists of two main procedures. (i) Fast boundary clustering algorithm decomposes scene features into multilayers, so that it is faster and easier to analyze Thai text characters. (ii) Modified connected component analysis method is applied to such scene features in order to detect Thai text boundaries. Based on 150 test images with 4,920 characters, the experimental results demonstrate that the proposed approach achieves the high average precision and recall, 0.80 and 0.90.

Speech recognition of different sampling rates using fractal code descriptor

Currently, the use of speech recognition is increaseingly in many applications such as mobile device interaction, interactive voice response system, voice search, voice dictation and voice identification. The heart of such applications is speech features needed to represent input signals. However, in real applications, speech signals are sampled with various sampling rates. The different sampling rates of input speech lead to the different features. This makes the speech recognition rate dropping. Therefore, this paper proposes an independent resolution descriptor based on fractal codes obtained by fractal encoding and decoding processes. The encoding process extracts fractal codes from partitioned speech signals, whereas the decoding process reconstructs independent resolution speech signals from the fractal codes. This method can effectively reconstruct speech signals at any sampling rates, especially at a higher sampling rate, which is a grand challenge. The proposed method is evaluated the performance by testing with AN4 corpus of CMU Sphinx speech recognition engine. The experimental results show that the proposed method can improve the accuracy of speech recognition, even if the sampling rate of testing speeches differs from that of training speeches.

Printed Thai Character Recognition Using Standard Descriptor

The various font-types, font-sizes, and font-styles have a great impact on recognition performance of optical character recognition (OCR) systems. This becomes a grand challenge for recognition improvement. In order to enhance the performance, this paper proposes the printed Thai character recognition using a standard descriptor. The descriptor construction consists of two principal phases—preprocessing and feature extraction. In the former phase, the preprocessing provides a standard form for each character image. In the latter phase, the singular value decomposition (SVD) is applied to all font-type, fontsize, and font-style character images to extract features. Then the standard descriptor is constructed from the suitable order selection of the SVD feature decomposition. Finally, the projection matrix technique is applied to the recognition phase in order to measure the cosine similarity between the standard descriptor and test set. The experimental results show that the proposed method achieves a high recognition rate and is invariant to font-types, font-sizes, and font-styles.

Offline handwritten signature recognition using polar-scale normalization

Offline handwritten signature is still widely used for person verification in financial and business transactions. Most research in offline handwritten signature at-tempts to improve feature extraction and classification for the better recognition rate. The deformation and unsteadiness of handwritten signatures, such as direction, declination, and size, are also the key factors sensitive to recognition rate. Therefore, this paper focuses on the pre-processing phase, which is an alternative way to improve the accuracy and to make such factors stable. This study is based on the hypothesis; a table signature size is able to boost up the recognition rate. Therefore, polar-scale normalization (PSN) is proposed to scale signature size and make it stable. In this method, the signature images are transformed into the polar coordinate system consisting of polar distance and angle, and then normalized by ‖norm‖. The normalized distance is certainly estimated by polar coordinate that helps reduce the deformed images. The 5,739-sample signature images with 150 classes are used to test in the experiment. PSN provide the better performance, when compared with traditional normalization schemes including min-max, decimal, z-score and MAD normalizations. The results reveal that the proposed method can improve the average recognition rate up to 98.39%.

An improved 2DPCA for face recognition under illumination effects

Principal component analysis (PCA) is one of the successful techniques for applying to face recognition, but its challenge still remains for solving an illumination effect condition. This paper proposes an improved 2DPCA (I-2DPCA) for overwhelming the illumination effect in face recognition. The proposed method is based on two assumptions. The first assumption is to create the covariance matrix that can effectively decompose the components of illumination effects from the eigenfaces. This avoids the illumination effect problem. The second assumption is to select the suitable eigenvectors that can significantly improve the recognition rate. Based on the Extended Yale Face Database B+ containing 60 illumination conditions, the experimental results show that not only does the proposed method decrease the computing time, but it also improves the recognition rate up to 95.93%.

Modified differential box-counting method using weighted triangle-box partition

Differential box-counting (DBC) is one of the commonly used methods to estimate fractal dimension (FD) for gray scale images. It has been successfully applied in many applications such as image segmentation, pattern recognition, texture analysis and medical signal analysis. However, the accuracy improvement of FD estimation is still a grand challenge. This paper proposes a modified differential box-counting method using weighted triangle-box partition (MDBC) to reduce the estimation error caused by an undercounting problem. The proposed method is derived from two assumptions: (i) increasing the precision of box-counts by using unequally triangle box partition, and (ii) weighting the box-counts in proportion to the size of triangle-box partition. Based on these assumptions, on each grid a square box is divided into four asymmetric triangle-box patterns. Each pattern is calculated the box-counts by a weighted box-counting technique. The maximum number of box-counts represents the better estimation. In this way, the experimental results show that MDBC outperforms the baseline methods in terms of fitting error. Furthermore, the proposed method applies to finger-knuckle-print recognition in order to test its efficiency. The results illustrate that it significantly enhances the recognition rate when compared with the conventional differential box-counting (DBC) and improved triangle box-counting in combination with DBC (ITBC-DBC) methods.

Offline handwritten signature recognition using adaptive variance reduction

Although offline handwritten signature recognition has been continually researched, it still requires an improvement of recognition rate. Most of existing techniques focus on feature extraction to improve their performance. This paper proposes an alternative way to increase the recognition rate by analyzing an important characteristic of input information, namely variability of signatures. The proposed method is based on the hypothesis; reducing the variability of signatures leads to boost up the recognition rate. Therefore, the variance reduction technique is applied to normalize offline handwritten signatures by means of an adaptive dilation operator. Then the variability of signatures is analyzed in terms of coefficient of variation (CV). The optimal CV is obtained and used to be a threshold limit value for the acceptable variance reduction. Based on 5,739 signature samples with 140 classes, the experimental results show that the adaptive variance reduction procedure helps improve the recognition rate when compared to the traditional schemes without adaptive variance reduction, including histogram of gradient (HOG) and pyramid histogram of gradient (PHOG) techniques.

Modified adaptive thresholding using integral image

Adaptive thresholding, the simple way to perform image segmentation, is a form of image thresholding used to classify pixels as dark and light. Taking grayscale image as an input for this task is only good in case that text appears in low intensity area. In case that text appears in high intensity area, it leads to lower recall rate for text detection process. However, it can be fixed by taking an inverted-grayscale image instead. The problem is how to determine automatically whether it is better to take the normal-grayscale or inverted-grayscale image for each original image. This paper proposes the simple way to do that by means of the adaptive thresholding using the integral image itself with some additional steps based on its principle. The proposed method consists of two main process, low/high intensity area segmentation and modified adaptive thresholding.

Hinge Loss Projection for Classification

Hinge loss is one-sided function which gives optimal solution than that of squared error SE loss function in case of classification. It allows data points which have a value greater than 1 and less than